5
$\begingroup$

Is there a Whitney sum formula for topological rational Pontryagin classes? I thought the answer is yes, but now I cannot find a reference. Is it even true? The PL case would also be of interest.

$\endgroup$
  • 1
    $\begingroup$ Let p(E) denote the total (integral) Pontryagin class of a real bundle E. It's not too hard to show that p(E (+) F) = p(E) * p(F) mod 2-torsion. The 2-torsion term is stated in Theorem 1.6 of Brown's "The Cohomology of BSO_n and BO_n with Integer Coefficients". $\endgroup$ – skd May 22 at 4:23
  • 1
    $\begingroup$ @sdk: Brown's paper is about vector bundles, while my question is about locally trivial bundles with fiber homeomorphic to $\mathbb R^n$. $\endgroup$ – Igor Belegradek May 22 at 12:40
  • $\begingroup$ Ah, yes. Sorry, didn't read carefully enough. $\endgroup$ – skd May 22 at 17:29
8
$\begingroup$

Yes. A simple argument is that $BO \to BTOP$ is a rational equivalence and an H-space map (in fact even an infinite loop map), so it follows from the Whitney sum formula for vector bundles.

Edit: The argument for this is as follows. Let $\mu : BTOP \times BTOP \to BTOP$ be the map corresponding to Whitney sum of (stable, topological) bundles. The question is whether the identity $$\mu^* p_n = \sum_{ a + b = n} p_a \otimes p_b$$ holds. As the map $BO \to BTOP$ is a rational equivalence and an H-space map, it is equivalent to verify this equation in the cohomology of $BO$ instead, where it indeed holds.

| cite | improve this answer | |
$\endgroup$
  • 1
    $\begingroup$ Thank you! For my record here is a reference: on p.215 of Boardman-Vogt's book "Homotopy invariant algebraic structures on topological spaces" the authors say (roughtly) that since $O\to Top\to F$ are topological monoid homomorphisms under the Whitney sum, these maps are infinite loop space maps, and hence so are the corresponding maps of classifying spaces $BO\to BTop\to BF$. Actually, instead of this last sequence they write $BO\to Top\to F$ which I assume is a misprint. $\endgroup$ – Igor Belegradek May 22 at 12:29
  • $\begingroup$ Is it true more generally that for any topological monoid (or at least group) homomorphism $G\to H$ the corresponding map of classifying spaces $BG\to BH$ is an infinite loop map? $\endgroup$ – Igor Belegradek May 22 at 12:29
  • $\begingroup$ @ConnorMalin: thank you, I see now that Boardman-Vogt's result has quite a few technical assumptions; they speak of symmetrical monoidal functors. I gather, the point is that Whitney sum is commutative. $\endgroup$ – Igor Belegradek May 22 at 13:05
  • $\begingroup$ Sorry, I deleted my comment in fear that I was missing something. For posterity, I said that usually a group will not be an infinite loop space. $\endgroup$ – Connor Malin May 22 at 13:35
  • 1
    $\begingroup$ $[X,BSO_\mathbb{Q}]\rightarrow [X,BSTOP_\mathbb{Q}]$ is an isomorphism. The rational Pontryagin classes of $x\in [X,BSTOP]$ only depend on its image in $[X,BSTOP_\mathbb{Q}]$. $\endgroup$ – archipelago May 23 at 12:52

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.